Hyaluronan regulates transforming growth factor-beta1 receptor compartmentalization.

Transforming growth factor-beta1 (TGF-beta1) is a key cytokine involved in the pathogenesis of fibrosis in many organs. We previously demonstrated in renal proximal tubular cells that the engagement of the extracellular polysaccharide hyaluronan with its receptor CD44 attenuated TGF-beta1 signaling. In the current study we examined the potential mechanism by which the interaction between hyaluronan (HA) and CD44 regulates TGF-beta receptor function. Affinity labeling of TGF-beta receptors demonstrated that in the unstimulated cells the majority of the receptor partitioned into EEA-1-associated non-lipid raft-associated membrane pools. In the presence of exogenous HA, the majority of the receptors partitioned into caveolin-1 lipid raft-associated pools. TGF-beta1 increased the association of activated/phosphorylated Smad proteins with EEA-1, consistent with activation of TGF-beta1 signaling following endosomal internalization. Following addition of HA, caveolin-1 associated with the inhibitory Smad protein Smad7, consistent with the raft pools mediating receptor turnover, which was facilitated by HA. Antagonism of TGF-beta1-dependent Smad signaling and the effect of HA on TGF-beta receptor associations were inhibited by depletion of membrane cholesterol using nystatin and augmented by inhibition of endocytosis. The effect of HA on TGF-beta receptor trafficking was inhibited by inhibition of HA-CD44 interactions, using blocking antibody to CD44 or inhibition of MAP kinase activation. In conclusion, we have proposed a model by which HA engagement of CD44 leads to MAP kinase-dependent increased trafficking of TGF-beta receptors to lipid raft-associated pools, which facilitates increased receptor turnover and attenuation of TGF-beta1-dependent alteration in proximal tubular cell function.

Transforming growth factor-␤1 (TGF-␤1) is a key cytokine involved in the pathogenesis of fibrosis in many organs. We previously demonstrated in renal proximal tubular cells that the engagement of the extracellular polysaccharide hyaluronan with its receptor CD44 attenuated TGF-␤1 signaling. In the current study we examined the potential mechanism by which the interaction between hyaluronan (HA) and CD44 regulates TGF-␤ receptor function. Affinity labeling of TGF-␤ receptors demonstrated that in the unstimulated cells the majority of the receptor partitioned into EEA-1-associated non-lipid raft-associated membrane pools. In the presence of exogenous HA, the majority of the receptors partitioned into caveolin-1 lipid raft-associated pools. TGF-␤1 increased the association of activated/phosphorylated Smad proteins with EEA-1, consistent with activation of TGF-␤1 signaling following endosomal internalization. Following addition of HA, caveolin-1 associated with the inhibitory Smad protein Smad7, consistent with the raft pools mediating receptor turnover, which was facilitated by HA. Antagonism of TGF-␤1-dependent Smad signaling and the effect of HA on TGF-␤ receptor associations were inhibited by depletion of membrane cholesterol using nystatin and augmented by inhibition of endocytosis. The effect of HA on TGF-␤ receptor trafficking was inhibited by inhibition of HA-CD44 interactions, using blocking antibody to CD44 or inhibition of MAP kinase activation. In conclusion, we have proposed a model by which HA engagement of CD44 leads to MAP kinase-dependent increased trafficking of TGF-␤ receptors to lipid raft-associated pools, which facilitates increased receptor turnover and attenuation of TGF-␤1-dependent alteration in proximal tubular cell function.
Transforming growth factor-␤1 (TGF-␤1) 1 is a multifunctional cytokine that is involved in maintenance of normal homeostasis. It is important in development and tissue differentiation and normal wound healing through its effects on cell proliferation, migration, differentiation, and apoptosis (1). In addition, it is a critical regulator of the normal inflammatory response as illustrated by the death of TGF-␤1 Ϫ/Ϫ mice of a multifocal inflammatory syndrome soon after weaning (2). Its growth inhibitory effect explains its role as a tumor suppressor, although its expression by tumor cells contributes to cancer progression and metastasis at later stages of disease (3,4). TGF-␤1 is also important in the pathological process of fibrosis and subsequent organ failure, leading to the concept of fibrotic disease as "the dark side of tissue repair" (5,6). These different actions of TGF-␤1 are related to altered epithelial cell responses to it in different cellular contexts (7,8).
Sustained production of TGF-␤1 is a key determinant of tissue fibrosis (6), including renal tubulointerstitial fibrosis (9). In chronic renal disease the severity of tubulointerstitial fibrosis is an excellent predictor of progression to end-stage renal failure (10,11). Recent studies demonstrate the expression of three isoforms of TGF-␤ in the kidney (TGF-␤1, 2, and 3), and although all three isoforms alter mesangial, interstitial fibroblast, and proximal tubular epithelial cell function the effects of TGF-␤2 and TGF-␤3 are mediated at least in part by stimulation of TGF-␤1 protein synthesis (12). Much of our work has focused on the mechanisms that regulate TGF-␤1 generation and function in the kidney and, more specifically, in renal proximal tubular epithelial cells.
TGF-␤s elicit their signaling effects by binding mainly to three cell-surface receptors: types I (RI), II (RII), and III (RIII). RI and RII are serine/threonine kinases that form heteromeric complexes and are necessary for TGF-␤ signaling, which is initiated when the ligand induces assembly of a heteromeric complex of type II and type I receptors. The RII kinase then phosphorylates RI on a conserved glycine-serine-rich domain. This activates the RI kinase, which subsequently recognizes and phosphorylates members of the intracellular receptor-regulated Smads (R-Smads) signal transduction pathway. For TGF-␤1 these include Smad2 and 3. This causes dissociation of the R-Smads from the receptor, stimulates the assembly of a heteromeric complex between the phosphorylated R-Smad and the Co-Smad Smad4, and induces the nuclear accumulation of this heteromeric Smad complex (reviewed in Ref. 13).
Endocytosis of cell surface receptors is an important regulatory event in signal transduction. TGF-␤1 receptors internalize into both caveolin-and EEA-1-positive vesicles and reside in both lipid raft and non-raft membrane domains (14). Clathrindependent internalization into the EEA1-positive endosome promotes TGF-␤1 signaling. In contrast, the lipid raft-caveolar internalization pathway contains Smad7-bound receptor and is required for receptor turnover. Segregation of TGF-␤1 receptors into distinct compartments, therefore, regulates TGF-␤1 receptor signaling and turnover.
Many studies have demonstrated an association between alteration in the generation of the extracellular matrix polysaccharide, hyaluronan (HA), and renal injury diseases (15)(16)(17)(18). The functional significance of this, however, is not clear. HA promotes the signaling interaction between the principle cell surface recep-* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ Supported by a senior fellowship from GlaxoSmithKline. To whom correspondence should be addressed. Tel.: 44-2920-748411; Fax: 44-2920-748470; E-mail: PhillipsAO@cf.ac.uk. 1 The abbreviations used are: TGF, transforming growth factor; MAP, mitogen-activated protein; RI, II and III, receptor types I, II, III; HA, hyaluronan; PTC, proximal tubular cell; MEK, mitogen-activated protein kinase/extracellular signal-regulated kinase kinase; Cav-1, caveolin-1. tor for HA, CD44, and the TGF-␤ type I receptor in metastatic breast tumor cells (19). Our studies with renal proximal tubular cells (PTC) have recently demonstrated that co-localization of CD44 and TGF-␤ receptors facilitates modulation of both Smadand non-Smad-dependent TGF-␤1-mediated events by HA (20). This suggests that alteration of HA synthesis may represent an endogenous mechanism to limit renal injury. The aim of the current study was to examine the effect of HA on TGF-␤1 receptor compartmentalization in PTC. More specifically, we have examined the hypothesis that HA-mediated down-regulation of TGF-␤1 signaling was the result of increased segregation of TGF-␤1 receptors into a lipid raft-caveolar compartment away from the endosomal signaling compartment.
In all aspects of cell biology that we have previously studied, HK-2 cells respond in an identical fashion to primary cultures of human proximal tubular cells (22)(23)(24)(25). HK-2 cells are therefore a good model from which general conclusions can be drawn in terms of proximal tubular cell biology.
Transient Transfection-The Smad-responsive promoter (SBE) 4 -Lux (26) and constitutively active Alk5 (27) were gifts from Aristidis Moustakas. For transfection of the reporter construct, 80 ϫ 10 3 cells/well were seeded onto a 12-well plate (this density of cells produced an 80% confluence monolayer the following day). The next day cells were transfected with 0.5 g of the Smad-responsive promoter-luciferase construct (or the constitutively active Alk5 construct) using the mixed lipofection reagent FuGene 6 (Roche Applied Science) at a ratio of 1.5 l of FuGene to 0.5 g of DNA in serum-free and insulin-free medium. Transfection efficiency was monitored by co-transfection with a ␤-galactosidase reporter plasmid. 24 h after transfection, cells were stimulated with either TGF-␤1 or the combination of TGF-␤1 and HA. Following lysis of the cells in reporter lysis buffer (Promega), luciferase content was quantified by glow-type luminescence assay (Bright-Glo; Promega), and ␤-galactosidase activity was determined by commercial assay (Promega). Luciferase activity was normalized to ␤-galactosidase activity.
Detergent-free Purification of Lipid Raft-rich Membrane Fraction-HK-2 cells and ALK5-transfected HK-2 cells were grown to near confluence in 100-mm dishes and affinity-labeled as described above. After two washes with ice-cold phosphate-buffered saline, two confluent dishes were scraped into 2 ml of 500 mM sodium carbonate, pH 11.0. Homogenization was carried out with 10 strokes of a tight-fitting Dounce homogenizer followed by three 10-s bursts of a tissue homoge-nizer (Powergen 125; Fisher Scientific, Loughborough, Leicestershire, UK), followed by three 20-s bursts of Ultrasonic disintegrator, (Soniprep 150; Fisher Scientific) to disrupt cellular membranes as previously described (29). The homogenates were adjusted to 45% sucrose by addition of 2 ml of 90% sucrose prepared in MBS (2-(N-morpholino)ethanesulfonic acid-buffered saline, 25 mM 2-(N-morpholino) ethanesulfonic acid, pH 6.5, 0.15 M NaCl) and placed at the bottom of an ultracentrifuge tube. A discontinuous sucrose gradient (4 ml of 35% sucrose, 4 ml of 5% sucrose, both prepared in MBS) was formed above and centrifuged at 39,000 rpm for 16 -20 h in an SW40 TI rotor (Beckman Instruments, Palo Alto, CA). A light-scattering band was observed at the 3-35% sucrose interface. Twelve 1-ml fractions were collected from the top of the tubes, and a portion of each fraction was analyzed by SDS-PAGE. A 330-kDa band ( 125 I-labeled TGF-␤-receptor complex) was detectable on autoradiography, and this was quantified by densitometry (Chemi Doc; Bio Rad). Data are expressed as percentage of the total [ 125 I]-labeled TGF-␤-receptor complex for that experiment in each fraction.
Immunoprecipitation and Immunoblotting/Western Analysis-Briefly, confluent monolayers were washed once with cold phosphatebuffered saline (Invitrogen), scraped, and rinsed into 5 ml of cold phosphate-buffered saline. After centrifugation at 2,500 rpm for 10 min, cell pellets were extracted in buffer (150 mM NaCl, 50 mM Tris-Cl, 0.01% NaN 3 , 2 mM EDTA, 1 mM sodium orthovanadate, 10 g/ml leupeptin, 25 g/ml aprotinin) containing 1% Triton X-100 for 30 min on ice. Samples were centrifuged at 12,500 rpm for 30 min, and then the supernatant (Triton-soluble components, including membrane and cytosolic fraction) was transferred to a separate tube and kept at Ϫ70°C until use.
Immunoprecipitation was performed by standard methodologies. Briefly, cell protein samples (200 g) were precleared with 25 l of packed protein A-cross-linked 4% beaded agarose (Sigma) at 4°C overnight. The beads were removed by centrifugation (13,000 rpm, 10 min) and the supernatant collected. Primary antibody (2 g/ml) was added to the cleared supernatant and incubated at 4°C with constant mixing for 4 h. The immune complex was captured by the addition of packed agarose protein A beads (50 l) overnight at 4°C. Beads were washed with radioimmune precipitation assay buffer (50 mM Tris, 150 mM NaCl, 0.5% sodium deoxycholate, 10 mM MgCl 2 , 0.1% SDS, 1% Triton X-100). 30 l of sample buffer was then added prior to boiling for 5 min. Separation of the beads was achieved by centrifugation (13,000 ϫ g for 10 min) and the supernatant removed. Specificity of immunoprecipitation was confirmed by negative control reactions performed with either no primary antibody or IgG control.
Subsequently, samples were subject to immunoblot/Western analysis. Briefly, equal amounts of sample were prepared in SDS sample buffer (2% SDS, 10% v/v glycerol, 60 mM Tris, and 0.05% v/v mercaptoethanol) and boiled for 5 min prior to loading onto 10% SDS-PAGE. Electrophoresis was carried out under reducing conditions according to the procedure of Laemmli (30). After electrophoresis the separated proteins were transferred to a nitrocellulose membrane (Amersham Biosciences). The membrane was blocked with Tris-buffered saline containing 5% nonfat powdered milk for 1 h and then incubated with the primary antibody in Tris-buffered saline containing 5% nonfat powdered milk and 0.1% Tween 20 (Tris-buffered saline-Tween) overnight at 4°C. The blots were subsequently washed in Tris-buffered saline-Tween and then incubated with an appropriate horseradish peroxidaseconjugated secondary antibody (Sigma) in Tris-buffered saline-Tween. Proteins were visualized using enhanced chemiluminescence (Amersham Biosciences) according to the manufacturer's instructions.

HA Increases TGF-␤1 Receptor Segregation into Lipid Rafts,
Associated with Cav-1-Previous studies have demonstrated co-localization of TGF-␤ receptors into both caveolin-and EEA-1-containing compartments. Caveolin associates with cholesterol-and sphingolipid-rich membrane domains called lipid rafts (31). To examine alterations in receptor partitioning, we adopted two experimental approaches: affinity labeling of endogenous TGF-␤1 receptors on HK-2 cells with [ 125 I]TGF-␤1 and affinity labeling of HK-2 cells transiently transfected with the constitutively active TGF-␤ type I receptor Alk5 (TGF0␤-I/T204D). Subsequently, rafts were fractionated by sucrose density centrifugation (32). Following addition of [ 125 I]TGF-␤1 (250 pM), receptors were found in both the raft and non-raft fractions (Fig. 1A). Scanning densitometry of the results of three separate experiments confirmed that following addition of [ 125 I]TGF-␤1, 66% of the total TGF-␤ receptor partitioned into the non-raft fractions (Fig. 1B). Western blot analysis with antibodies to either caveolin-1 (Cav-1) or EEA-1 confirmed partitioning of lipid rafts because the presumptive raft fractions contained Cav-1 and the non-raft fractions contained EEA-1 (Fig. 1C).
In contrast, addition of [ 125 I]TGF-␤1 in the presence of 25 g/ml HA led to a significant increase in TGF-␤ receptor detected in the raft fraction. Under these conditions, 58% of total TGF-␤ receptor partitioned into the raft fractions (mean of n ϭ 3, p ϭ 0.003). Disruption of cholesterol by pretreatment of cells with 50 g/ml nystatin (31) at 37°C for 1 h prevented a HAmediated shift of TGF-␤ receptor into the raft fraction (Fig. 1A). Following nystatin pretreatment and addition of [ 125 I]TGF-␤1 in the presence of 25 g/ml HA, only 39% of total TGF-␤ receptor was found in the raft fractions (Fig. 1C). This was not significantly different from that seen following addition of [ 125 I]TGF-␤1 alone. The effect of nystatin was dependent on cholesterol sequestration because it was prevented by addition of cholesterol (25 g/ml) together with nystatin to cells at 37°C for 1 h prior to addition of TGF-␤1 together with HA. This resulted in 55% of the total TGF-␤ receptor partitioning into the raft fractions. Statistically this was no different from the amount of receptor detected in the raft fraction when [ 125 I]TGF-␤1 together with HA were added to cells in the absence of nystatin.
Previously we have demonstrated by immunoprecipitation of CD44 that both types I and II TGF-␤1 receptors associate with CD44 (20). In metastatic breast tumor cells, however, the association was predominantly with the type I receptor. Because TGF-␤1 receptor heterodimers form upon receptor ligand interaction, we sought to clarify the nature of CD44-TGF-␤1 trafficking by using a constitutively active TGF-␤1 type I receptor construct. HK-2 cells were transfected with the Alk5 construct and TGF-␤ receptors affinity-labeled by the addition of [ 125 I]TGF-␤1 in the presence or absence of exogenous HA. As with endogenous TGF-␤ receptors, trafficking of transfected constitutively active type I receptor Alk5 was altered by the presence of HA ( Fig. 2A). In transiently transfected cells, in the presence of 250 pM [ 125 I]TGF-␤1 alone, 30% of total receptor was detected in the raft fractions. When cells were incubated with [ 125 I]TGF-␤1 together with HA (25 mg/ml), this increased to 62% (Fig. 2B). Disruption of cholesterol by pretreatment with nystatin also prevented an HA-dependent increase in receptor segregation into the raft fractions. Pretreatment of cells with nystatin prior to addition of [ 125 I]TGF-␤1 together with HA resulted in only 31% of the total TGF-␤ receptor being associated with the raft fractions, which was no different from that seen when [ 125 I]TGF-␤1 was added in the absence of HA. As with expression of endogenous receptors, the effect of nystatin was dependent on cholesterol sequestration because it was prevented by addition of cholesterol (25 g/ml) together with nystatin to cells at 37°C for 1 h prior to addition of TGF-␤1 together with HA. This resulted in 52% of the total TGF-␤ receptor partitioning into the raft fractions. Statistically, this was no different from the amount of receptor detected in the raft fraction when [ 125 I]TGF-␤1 together with HA were added to cells in the absence of nystatin.
Alterations in TGF-␤ Type 1 Receptor Associations-To determine the functional consequence of alteration in receptor partitioning, we stimulated cells with TGF-␤1 (10 ng/ml) in the presence and absence of HA (25 g/ml). Subsequently, we examined the association of TGF-␤ type I receptor with either EEA-1 or Cav-1 by immunoprecipitation of the TGF-␤ receptor and Western analysis of the latter two proteins.
Stimulation of cells with recombinant TGF-␤1 (10 ng/ml) and immunoprecipitation of TGF-␤ type I receptor led to an increase in association of EEA-1 with the receptor (Fig. 3A). This increase in association of TGF-␤1 type I receptor with EEA-1 following addition of recombinant TGF-␤1 was not seen when cells were stimulated by recombinant TGF-␤1 in the presence of HA (25 g/ml). Scanning densitometry of three separate experiments confirmed a statistically significant increase in association of EEA-1 with TGF-␤ type I receptor, which was abrogated by the presence of HA (Fig. 3B). The effect of HA on trafficking of TGF-␤ receptors was assessed by comparison of distribution in HK-2 cells exposed to either 250 pM radiolabeled TGF-␤1 (T) alone or radiolabeled TGF-␤1 in the presence of 25 g/ml HA molecular weight 2 ϫ 10 6 (TϩHA). In all experiments subcellular fractionation was performed 24 h after addition of the stimuli. To disrupt lipid rafts cells were pretreated with 50 g/ml nystatin at 37°C for 1 h prior to the addition of radiolabeled TGF-␤1 together with HA for a further 24 h (TϩHAϩN). Further characterization of the contribution of lipid rafts was examined by pretreatment of cells with 50 g/ml nystatin for 1 h at 37°C prior to the addition of radiolabeled TGF-␤1 together with HA in the presence of 25 g/l cholesterol for a further 24 h (TϩHAϩNϩC). An equal volume from each fraction was analyzed by SDS-PAGE electrophoresis followed by autoradiography. B, following scanning densitometry of autoradiographs, the distribution of TGF-␤ receptor into the raft (fractions 5 and 6) and non-raft (fractions 7-10) was quantified and the data from three separate experiments expressed graphically. C, to confirm partitioning of fractions, in parallel experiments equal volumes of each fraction following SDS-PAGE electrophoresis were subjected to Western blotting with anti-caveolin (cav-1) or anti-EEA1 antibodies.
In contrast, immunoprecipitation of TGF-␤ type I receptor and Western analysis of Cav-1 demonstrated that addition of TGF-␤1 (10 ng/ml) decreased the association of the two proteins (Fig. 3A). HA (25 g/ml) alone increased association of TGF-␤ type I receptor and Cav-1. Furthermore, when cells were stimulated with TGF-␤1 in the presence of HA, increased association of TGF-␤ type 1 receptor and Cav-1 was also apparent. Scanning densitometry confirmed the statistical significance of these changes (Fig. 3C).
Previously, we have demonstrated that HA-dependent inhibition of TGF-␤1 signaling was inhibited by co-incubation with a blocking antibody to CD44. Furthermore, we have demonstrated that addition of HA led to CD44-mediated activation of MEK activity and that inhibition of MEK by the inhibitor PD98059 also prevented the effects of HA on TGF-␤1 signaling. In keeping with these observations, HA-mediated decreased association of EEA-1 with TGF-␤ type I receptor following addition of TGF-␤1 was prevented by co-incubation with either blocking antibody to CD44 or the inhibitor of MEK, PD98059 (Fig. 3, A and B). Similarly co-incubation of HA with blocking antibody to CD44 or PD98059 and stimulation with recombinant TGF-␤1 prevented the increased association of TGF-␤ type I receptor and Cav-1 seen when cells were stimulated with TGF-␤1 in the presence of HA in the absence of either the blocking antibody to CD44 or PD98059 (Fig. 3, A and C). Further determination of the functional significance of alteration in TGF-␤1 type I receptor segregation was examined by defining the association between EEA-1 and phosphorylated Smad2/3 and the association between Cav-1 and the inhibitory Smad, Smad7 (Fig. 4).
Association of EEA-1 with phosphorylated Smad2/3 was determined by immunoprecipitation of EEA-1 and Western analysis of phosphorylated Smad2/3 (Fig. 4A). Following addition of TGF-␤1 alone there was a significant increase in the association of EEA-1 with phosphorylated Smad2/3 (Fig. 4B). In contrast, stimulation of cells with TGF-␤1 (10 ng/ml) in the presence of HA (25 g/ml) prevented the increased association of EEA-1 and phosphorylated Smads. This effect of HA on EEA-1 association with phosphorylated Smads was prevented by incubation with either blocking antibody to CD44 or incubation with the MEK inhibitor, PD98059.
TGF-␤1 stimulation followed by immunoprecipitation of Smad7 and Western analysis of associated Cav-1 demonstrated that addition of TGF-␤1 alone led to a decrease in association of Smad7 with Cav-1, whereas the addition of TGF-␤1 in the presence of HA increased the association of Smad7 with the inhibitory Cav-1 (Fig. 4, A and C). When these experiments were performed in the presence of either blocking antibody to FIG. 2. HA-mediated alteration in distribution of transfected Alk5. HK-2 cells transiently transfected with the constitutively active TGF-␤ type I receptor, Alk5, were affinity labeled with [ 125 I]TGF-␤1 and subjected to sucrose gradient subcellular fractionation to separate lipid rafts from other cellular components. The effect of HA on trafficking of TGF-␤ receptors was assessed by comparison of distribution in transfected cells exposed to either 250 pM radiolabeled TGF-␤1 (T) alone or radiolabeled TGF-␤1 in the presence of 25 g/ml HA molecular weight 2 ϫ 10 6 (TϩHA). Subcellular fractionation was performed 24 h after addition of the stimuli in all experiments. To disrupt lipid rafts, cells were pretreated with 50 g/ml nystatin at 37°C for 1 h prior to the addition of radiolabeled TGF-␤1 together with HA for a further 24 h (TϩHAϩN). The contribution of lipid rafts was further examined by pretreatment of cells with 50 g/ml nystatin for 1 h at 37°C prior to the addition of radiolabeled TGF-␤1 together with HA in the presence of 25 g/liter of cholesterol for a further 24 h (TϩHAϩNϩC). An equal volume from each fraction was analyzed by SDS-PAGE electrophoresis followed by autoradiography. B, following scanning densitometry of autoradiographs, the distribution of TGF-␤ receptor into the raft (fractions 5 and 6) and non-raft (fractions 7-10) was quantified and the data from three separate experiments expressed graphically.

FIG. 3. Alteration in association of TGF-␤ receptor with EEA-1 and Cav-1.
Confluent growth-arrested HK-2 cells were stimulated with recombinant TGF-␤1 or HA for 24 h as indicated in the presence or absence of either blocking antibody to CD44 (␣CD44) or the MEK inhibitor, PD98059 (MEKi). Subsequently, TGF-␤ type I receptor was immunoprecipitated as described under "Materials and Methods" and EEA-1 or Cav-1 expression in the precipitate examined by Western analysis (A). Following scanning densitometry, alteration in EEA-1 (B) or Cav-1 (C) was expressed as fold increase in densitometric ratio compared with control. The data represent the mean Ϯ S.D. of three separate experiments.
CD44 or the MEK inhibitor PD98059, there was no increase in the association of Smad7 with Cav-1 following stimulation with TGF-␤1 in the presence of HA.
Effect of Inhibition of TGF-␤1 Receptor Trafficking-Previous studies have demonstrated that cholesterol depletion by the addition of nystatin shifted receptors into non-raft compartments. The results presented above also demonstrated that pretreatment of cells with nystatin prevented a HA-mediated shift of TGF-␤ receptor into the raft fraction. In the current study we determined the effect of nystatin on attenuation of TGF-␤1-stimulated activation of the Smad signaling pathway using the (SBE) 4 -Lux reporter, which contains four repeats of the CAGACA sequence identified as a Smad binding element. Addition of TGF-␤1 led to a 10-fold increase in luciferase activity of the reporter construct (Fig. 5). Addition of HA did not increase the signal above control values, but addition of HA in the presence of TGF-␤1 led to a significant decrease in luciferase activity (Fig. 5). This represented a 27% reduction in luciferase activity as compared with stimulation with TGF-␤1 alone (mean of n ϭ 6, p Ͻ 0.005). In contrast, disruption of cholesterol by pretreatment of cells with nystatin (31) at 37°C for 1 h led to a dose-dependent reversal in HA-mediated attenuation of TGF-␤1 stimulation of luciferase activity, such that there was no difference in luciferase activity following addition of TGF-␤1 alone and TGF-␤1 in combination with HA. This effect of nystatin was dependent on chelation of cholesterol because addition of cholesterol (25 g/ml) together with nystatin to cells at 37°C for 1 h prior to addition of TGF-␤1 together with HA restored TGF-␤1-dependent Smad signaling.
Next we examined the effect of inhibition of clathrin-mediated endocytosis by K ϩ depletion, which prevents clathrin lattice assembly and has been shown to inhibit endosomal-dependent TGF-␤1 signaling (14). Activation of the TGF-␤1 signaling pathway was assessed by using the (SBE) 4 -Lux reporter (Fig. 5B). TGF-␤1-stimulated increase in luciferase activity was significantly attenuated when carried out in minimal medium. In addition, attenuation of TGF-␤1 signaling by the addition of HA was enhanced when carried out in the presence of minimal medium and was significantly greater than the effect of potassium depletion by addition of minimal medium alone. DISCUSSION TGF-␤1, which is the prototypic member of the TGF-␤ superfamily, exerts a broad range of biological activities. It plays pivotal roles during embryonic development, where it is involved in induction of cell differentiation and organogenesis. TGF-␤1 has been implicated in the pathogenesis of renal fibrosis in both experimental and human disease (33)(34)(35)(36)(37). A major function of TGF-␤1 is to regulate the expression of genes, the products of which contribute to the formation and degradation of extracellular matrix (ECM) (38 -42). Generally, TGF-␤1 leads to the accumulation of ECM by decreasing the synthesis of proteases and by increasing the levels of protease inhibitors (43). It also increases the expression of integrins through which ECM proteins such as fibronectin and collagen interact with cells (44,45). In vitro studies also suggest that TGF-␤1 induces phenotypic alterations in PTC, using intermediate filament markers and reorganization of the cytoskeleton with cells as indicators of a "fibroblastic" phenotype (46). Studies utilizing normal rat PTC also suggest that TGF-␤1 is a key mediator regulating differentiation of PTC into ␣-smooth muscle actinpositive cells (47). Not only is there strong evidence that TGF-␤1 is a key mediator of progressive renal fibrosis but attenuation of its action has been postulated to be a target for therapeutic intervention in numerous disease models (34,35,48,49). Understanding the mechanisms that regulate TGF-␤1dependent responses is therefore an important goal.
We previously demonstrated that the association of CD44 and TGF-␤1 receptors facilitated attenuation of PTC response to TGF-␤1 (20). More specifically, we demonstrated a decrease in synthesis of collagen in response to TGF-␤1 and decreased nuclear translocation of Smad4 when cells were stimulated with TGF-␤1 in the presence of HA. In addition to HA antagonism of TGF-␤1 extracellular matrix generation, HA antagonized the effect of TGF-␤1 on PTC migration. In contrast to the effect of TGF-␤1 on collagen synthesis, which is Smad-dependent, the anti-migratory effect of TGF-␤1 in this model is known to be dependent on activation of RhoA (50). In the presence of HA, TGF-␤1-mediated activation of RhoA was also abrogated in a CD44-dependent manner (20). This suggests that co-localization of CD44 and TGF-␤ receptors facilitates modulation of both Smad-and non-Smad-dependent TGF-␤1-mediated FIG. 4. Phosphorylated Smads and inhibitory Smads localize to distinct subcellular fraction. Confluent growtharrested HK-2 cells were stimulated with recombinant TGF-␤1 or HA for 24 h as indicated in the presence or absence of either blocking antibody to CD44 (␣CD44) or the MEK inhibitor PD98059 (MEKi). Subsequently, either EEA-1 or Smad7 were immunoprecipitated as described under "Materials and Methods" and phospho-Smad2/3 or Cav-1 expression in the precipitate examined by Western analysis (A). Following scanning densitometry, alteration in phospho-Smad2/3 (B) or Cav-1 (C) was expressed as fold increase in densitometric ratio compared with control. The data represent the mean Ϯ S.D. of three separate experiments. events by HA. Interestingly, this effect of HA was only seen with HA of a high molecular weight (2 ϫ 10 6 ), whereas HA of much lower molecular weight (65,000) did not antagonize the effect of TGF-␤1 (20). CD44 binds HA of high and low molecular weight and in doing so has distinct function. Antagonism of TGF-␤1-mediated effects by high molecular weight HA is consistent with the assumption that, in general, high molecular weight HA represents the normal homeostatic state, whereas the generation of low molecular weight HA fragments signals a disruption of the normal homeostatic environment. In terms of renal disease, increased expression of HA is known to occur in both acute and chronic models of injury, and we have proposed that its role is to facilitate repair and limit progressive fibrotic effects that underlie progressive renal dysfunction. The aim of the work outlined in the current study was therefore to determine the mechanism by which HA-CD44 interactions downregulate TGF-␤1-dependent events.
Clathrin-coated pit-mediated endocytosis is traditionally considered a major mechanism by which cells regulate the level of cell surface receptors. For TGF-␤1, however, Smad2 or -3 activation and downstream signaling occurs after endocytic vesicle formation (51). Thus, TGF-␤ receptor endocytosis is required to propagate post-receptor signaling events. Endocytosis via clathrin-coated pits of TGF-␤ receptor is constitutive, occurring in the absence of ligand (52). It has been demonstrated that this process is dependent on a short sequence (residues Ile-218-Ile-219-Leu-220) that conforms to the dileucine family of internalization signals.
In mammalian cells, internalization of cell surface proteins occurs through both clathrin-dependent and -independent pathways. Recent studies have demonstrated TGF-␤ Ser-Thr kinase receptor internalization through the classical clathrin-dependent pathway and also through the raft-caveolin route, with the former pathway facilitating Smad activation and the latter mediating receptor degradation (14). Furthermore, regulation of TGF-␤ receptor trafficking was not regulated by ligand stimulation, suggesting that the kinase activity of the receptors themselves, although regulating signal transduction, do not regulate receptor trafficking. Although the functional significance of two TGF-␤ receptor pools is now apparent, very little is known regarding the regulation of trafficking of TGF-␤ receptors between the non-raftsignaling and raft-associated degradative pools. The data in the current study demonstrate that interaction of HA with its receptor CD44 leads to an increase in the trafficking of TGF-␤ receptor to lipid raft-associated membrane pools. Furthermore, this results in an increase in the association of the TGF-␤ receptor with Smad7. Furthermore, in the presence of HA, concomitant with TGF-␤ receptor partitioning into the lipid raft compartment, there was a decrease in association of activated phosphorylated Smad protein with it. This suggests that HA-dependent attenuation of TGF-␤1 signaling is related to increased receptor turnover and decreased endosomal internalization. This regulation of TGF-␤1 signaling at the level of the TGF-␤1 receptor itself is consistent with our previous observation that HA attenuated both Smad-and non-Smad-dependent TGF-␤1 signaling events.
In our previous studies, antagonism of TGF-␤1-Smad-mediated events such as increased collagen synthesis was prevented by inhibition of MAP kinase (20). We speculated that this may be related to CD44-mediated activation of the MAP kinase cascade and subsequent phosphorylation of TGF-␤1-regulated Smad proteins, because it is well established that phosphorylation of MAP kinase sites in the linker region of Smad proteins are known to negatively regulate their function (53). It has been previously demonstrated that the cytoplasmic domain of CD44 binds to the TGF-␤ type I receptor at a single site with high affinity (19). Rather than affecting Smad protein linker region phosphorylation, it seems more likely that activation of MAP kinase following engagement of CD44 by HA facilitates CD44-TGF-␤ type I receptor interaction and subsequent trafficking of TGF-␤ receptors to lipid raft fractions. This is supported by our previous data demonstrating inhibition of the functional consequences of HA on TGF-␤1 collagen generation and cell migration by either a blocking antibody to CD44 or inhibition of MEK activity using PD98059 (20). In the current study we have also demonstrated that a blocking antibody to CD44 or PD98059 also prevented HA-mediated trafficking of TGF-␤ receptor to lipid rafts and their association with both Cav-1 and Smad7. The current data would suggest that down- FIG. 5. Attenuation of TGF-␤1 receptor signaling. A, sequestration of TGF-␤1 from lipid rafts. HK-2 cells were transiently transfected with the Smad-responsive promoter (SBE) 4 -lux prior to stimulation with recombinant TGF-␤1 (10 ng/ml) either in the presence or absence of HA (25 g/ml, molecular weight 2 ϫ 10 6 ) for 24 h. The role of lipid rafts in Smad signaling was examined by pretreatment of transfected cells with 50 g/ml nystatin at 37°C for 1 h or pretreatment with 50 g/ml nystatin at 37°C for 1 h prior to the addition of TGF-␤1 together with HA in the presence of 25 g/ml cholesterol. Subsequently, luciferase content was quantified as described under "Materials and Methods" and the results normalized for transfection efficiency (using ␤-galactosidase) expressed as the fold increase above the non-stimulated control. The data represent the mean Ϯ S.D. of six individual experiments. B, inhibition of endosomal internalization. HK-2 cells transfected with the reporter construct were incubated in medium (Dulbecco's modified Eagle's medium/Ham's F12):water (1:1) for 5 min at 37°C followed by incubation in minimal medium (serum-free medium containing 20 mM Hepes, pH 7.5, 140 mM sodium chloride, 1 mM calcium chloride, 1 mM magnesium sulfate, and 5.5 mM glucose) for 1 h at 37°C prior to stimulation with TGF-␤1 (10 ng/ml) either in the presence or absence of HA (25 g/ml, MW 2 ϫ 10 6 ) for 24 h. Stimulation was also carried out in the minimal medium. In control experiments, cells were exposed to minimal medium supplemented with 10 mM potassium chloride. Subsequently, luciferase content was quantified as described under "Materials and Methods" and the results normalized for transfection efficiency (using ␤-galactosidase) expressed as the fold increase above the non-stimulated control. The data represent the mean Ϯ S.D. of four individual experiments; *, p Ͻ0.05 compared with TGF-␤1 and potassium-supplemented minimal medium. regulation of TGF-␤1 signaling is not the result of modulation of Smad protein expression. It seems more likely that the role of MAP kinase activation is to regulate the trafficking of TGF-␤ receptors to lipid raft fractions. Furthermore, although we have previously demonstrated that both type I and type II receptors may bind to CD44, in the current study we demonstrated that addition of HA altered the trafficking of TGF-␤ receptors when cells were transfected with the type I receptor alone. This would suggest that formation of heterodimers of the type I and II receptors are not necessary to facilitate HA-mediated trafficking of the receptors to the lipid raft pool and that it is the regulation of the direct association of the type I receptor with CD44 that is critical for receptor trafficking.
Our results demonstrating attenuation of TGF-␤1 signaling by HA are in contrast to recent reports utilizing metastatic breast tumor cells in which HA increased CD44 interaction with the TGF-␤ receptor I kinase and increased SMAD2/ SMAD3 phosphorylation (19). It is clear that TGF-␤1 has both tumor suppressor and tumor promoting activity at different stages of tumorigenesis. This dual-or multifunctionality of TGF-␤1 has been the source of much research. Cell surface expression of certain CD44 isoforms is closely correlated to the development of numerous tumors and their prognosis. In the study utilizing the breast tumor cell line (MDA-MB-231), a single CD44 isoform CD44v3 was expressed (19). In contrast, we have previously demonstrated that numerous isoforms of CD44 are expressed in PTC (54). More specifically, in addition to expression of the standard form of CD44, human proximal tubular cells also expressed epithelial CD44 (CD44 v8-v10), pMeta 1(CD44 v4-v7), pMeta2 (CD44 v6,7) and keratinocyte CD44 (CD44 v3-v10). It is interesting to speculate that one mechanism by which TGF-␤1 responses may be regulated may therefore be related to the differential expression of CD44 isoforms, with some isoforms enhancing TGF-␤1 signaling and others attenuating TGF-␤1-dependent responses.
In conclusion, we have demonstrated that engagement of CD44 by HA in PTC attenuates TGF-␤1 signaling by increasing trafficking of TGF-␤ receptors to non-signaling lipid raft-associated pools, which is likely to increase receptor turnover.